The long downward trend in the practice of breast feeding was reversed during the 1972-73 period. Data from the National Survey of Family Growth conducted by the National Center for Health Statistics were used to investigate the social correlates of breast feeding during the periods 1970-72 and 1973-75 to determine if these factors were related to the reversal in the breast feeding trend. A multivariate log linear modeling technique was used to test hypotheses regarding the direct and indirect effects of education, race, employment status, and source of prenatal care. While education, race, and employment status were directly related to the breast feeding decision, the analysis showed that the trend in breast feeding was unrelated to these correlates. Two alternate conclusions may be drawn from these findings: first, it is possible that changes in infant feeding practices occur earlier in some groups than in others, but the characteristics that distinguish such groups are not included in conventional social demographic data. Alternately, it is possible that the practice of breast feeding appeals equally to all social groups, and changes in the practice occur in response to broad social forces which affect society as a whole.